feat(telemetry): clean up spans (#1760)

llama_stack/distribution/server/endpoints.py

@@ -5,6 +5,7 @@
 # the root directory of this source tree.
 
 import inspect
+import re
 from typing import Dict, List
 
 from pydantic import BaseModel
@@ -19,6 +20,7 @@ class ApiEndpoint(BaseModel):
     route: str
     method: str
     name: str
+    descriptive_name: str | None = None
 
 
 def toolgroup_protocol_map():
@@ -58,8 +60,69 @@ def get_all_api_endpoints() -> Dict[Api, List[ApiEndpoint]]:
                 method = "delete"
             else:
                 method = "post"
-            endpoints.append(ApiEndpoint(route=route, method=method, name=name))
+            endpoints.append(
+                ApiEndpoint(route=route, method=method, name=name, descriptive_name=webmethod.descriptive_name)
+            )
 
         apis[api] = endpoints
 
     return apis
+
+
+def initialize_endpoint_impls(impls):
+    endpoints = get_all_api_endpoints()
+    endpoint_impls = {}
+
+    def _convert_path_to_regex(path: str) -> str:
+        # Convert {param} to named capture groups
+        # handle {param:path} as well which allows for forward slashes in the param value
+        pattern = re.sub(
+            r"{(\w+)(?::path)?}",
+            lambda m: f"(?P<{m.group(1)}>{'[^/]+' if not m.group(0).endswith(':path') else '.+'})",
+            path,
+        )
+
+        return f"^{pattern}$"
+
+    for api, api_endpoints in endpoints.items():
+        if api not in impls:
+            continue
+        for endpoint in api_endpoints:
+            impl = impls[api]
+            func = getattr(impl, endpoint.name)
+            if endpoint.method not in endpoint_impls:
+                endpoint_impls[endpoint.method] = {}
+            endpoint_impls[endpoint.method][_convert_path_to_regex(endpoint.route)] = (
+                func,
+                endpoint.descriptive_name or endpoint.route,
+            )
+
+    return endpoint_impls
+
+
+def find_matching_endpoint(method, path, endpoint_impls):
+    """Find the matching endpoint implementation for a given method and path.
+
+    Args:
+        method: HTTP method (GET, POST, etc.)
+        path: URL path to match against
+        endpoint_impls: A dictionary of endpoint implementations
+
+    Returns:
+        A tuple of (endpoint_function, path_params, descriptive_name)
+
+    Raises:
+        ValueError: If no matching endpoint is found
+    """
+    impls = endpoint_impls.get(method.lower())
+    if not impls:
+        raise ValueError(f"No endpoint found for {path}")
+
+    for regex, (func, descriptive_name) in impls.items():
+        match = re.match(regex, path)
+        if match:
+            # Extract named groups from the regex match
+            path_params = match.groupdict()
+            return func, path_params, descriptive_name
+
+    raise ValueError(f"No endpoint found for {path}")

llama_stack/distribution/server/server.py

@@ -32,6 +32,10 @@ from llama_stack.distribution.request_headers import (
     request_provider_data_context,
 )
 from llama_stack.distribution.resolver import InvalidProviderError
+from llama_stack.distribution.server.endpoints import (
+    find_matching_endpoint,
+    initialize_endpoint_impls,
+)
 from llama_stack.distribution.stack import (
     construct_stack,
     redact_sensitive_fields,
@@ -222,20 +226,18 @@ def create_dynamic_typed_route(func: Any, method: str, route: str):
 
 
 class TracingMiddleware:
-    def __init__(self, app):
+    def __init__(self, app, impls):
         self.app = app
+        self.impls = impls
 
     async def __call__(self, scope, receive, send):
         if scope.get("type") == "lifespan":
             return await self.app(scope, receive, send)
 
         path = scope.get("path", "")
-
-        # Try to match the path to a route template
-        route_template = self._match_path(path)
-
-        # Use the matched template or original path
-        trace_path = route_template or path
+        if not hasattr(self, "endpoint_impls"):
+            self.endpoint_impls = initialize_endpoint_impls(self.impls)
+        _, _, trace_path = find_matching_endpoint(scope.get("method", "GET"), path, self.endpoint_impls)
 
         trace_context = await start_trace(trace_path, {"__location__": "server", "raw_path": path})
 
@@ -251,36 +253,6 @@ class TracingMiddleware:
         finally:
             await end_trace()
 
-    def _match_path(self, path):
-        """Match a path to a route template using simple segment matching."""
-        path_segments = path.split("/")
-
-        for route in self.app.app.routes:
-            if not hasattr(route, "path"):
-                continue
-
-            route_path = route.path
-            route_segments = route_path.split("/")
-
-            # Skip if number of segments doesn't match
-            if len(path_segments) != len(route_segments):
-                continue
-
-            matches = True
-            for path_seg, route_seg in zip(path_segments, route_segments, strict=True):
-                # If route segment is a parameter (contains {...}), it matches anything
-                if route_seg.startswith("{") and route_seg.endswith("}"):
-                    continue
-                # Otherwise, segments must match exactly
-                elif path_seg != route_seg:
-                    matches = False
-                    break
-
-            if matches:
-                return route_path
-
-        return None
-
 
 class ClientVersionMiddleware:
     def __init__(self, app):
@@ -399,7 +371,6 @@ def main():
     logger.info(yaml.dump(safe_config, indent=2))
 
     app = FastAPI(lifespan=lifespan)
-    app.add_middleware(TracingMiddleware)
     if not os.environ.get("LLAMA_STACK_DISABLE_VERSION_CHECK"):
         app.add_middleware(ClientVersionMiddleware)
 
@@ -463,6 +434,7 @@ def main():
     app.exception_handler(Exception)(global_exception_handler)
 
     app.__llama_stack_impls__ = impls
+    app.add_middleware(TracingMiddleware, impls=impls)
 
     import uvicorn